Phân tích dữ liệu bảng trên STATA (POOL, FEM, REM, GLS, GMM, ARDL, MG, PMG, PCSE, DFE, (Kiểm định Hausman, đa cộng tuyến, phương sai thay đổi, tự tương quan)

Phân tích dữ liệu bảng trên STATA (POOL, FEM, REM, GLS, GMM, ARDL, MG, PMG, PCSE, DFE, (Kiểm định Hausman, đa cộng tuyến, phương sai thay đổi, tự tương quan)Phân tích dữ liệu bảng trên STATA (POOL, FEM, REM, GLS GMM, ARDL, MG, PMG, PCSE, DFE, (Kiểm định Hausman, đa cộng tuyến, phương sai thay đổi, tự tương quan)Phân tích dữ liệu bảng trên STATA (POOL, FEM, REM, GLS GMM, ARDL, MG, PMG, PCSE, DFE, (Kiểm định Hausman, đa cộng tuyến, phương sai thay đổi, tự tương quan)Phân tích dữ liệu bảng trên STATA (POOL, FEM, REM, GLS GMM, ARDL, MG, PMG, PCSE, DFE, (Kiểm định Hausman, đa cộng tuyến, phương sai thay đổi, tự tương quan)Phân tích dữ liệu bảng trên STATA (POOL, FEM, REM, GLS GMM, ARDL, MG, PMG, PCSE, DFE, (Kiểm định Hausman, đa cộng tuyến, phương sai thay đổi, tự tương quan)

KẾT QUẢ MINH HỌA CÁC BƯỚC PHÂN TÍCH DATA PANEL (DỮ LIỆU BẢNG)

1 CÁC CÂU LỆNH NỀN TẢNG

encode countrycode,gen(code)

xtset code year

panel variable: code (strongly balanced)

time variable: year, 2011 to 2023

***TỪ CƠ BẢN ĐẾN NÂNG CAO***

***CÓ ĐẦY ĐỦ TÀI LIỆU & DỮ LIỆU***

***ĐA DẠNG CÁC LOẠI DỮ LIỆU***

4 KIỂM ĐỊNH SỰ PHỤ THUỘC CHÉO (Dành cho Dữ liệu Bảng Panel Data)

a + 16.984 0.000 13.00 + 0.06 0.58 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

Ta thấy giá trị P-Value = 0.000 = 0.00% < 5%: Có tồn tại sự phụ thuộc chéo

Các biến dưới đây giải thích tương tự

b + 179.371 0.000 13.00 + 0.62 0.75 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

c + 155.706 0.000 13.00 + 0.54 0.89 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

d + 32.449 0.000 13.00 + 0.11 0.47 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

e + 100.277 0.000 13.00 + 0.35 0.64 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

f + 27.38 0.000 13.00 + 0.09 0.32 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

g + 116.053 0.000 13.00 + 0.40 0.53 | -+ Notes: Under the null hypothesis of cross-section independence, CD ~ N(0,1)

P-values close to zero indicate data are correlated across panel groups

Tham khảo cách kiểm định sự phụ thuộc chéo (Dành cho Dữ liệu chuỗi thời gian Time series Data)

xtreg a b c d e f g

Random-effects GLS regression Number of obs = 1,482

Group variable: code Number of groups = 114

R-sq: Obs per group:

Pesaran's test of cross sectional independence = 5.398, Pr = 0.0000

Câu lệnh kiểm định sự phụ thuộc chéo xtcsd,pesaran được thực hiện ngay sau kết quả

mô hình xtreg a b c d e f g

Kết quả kiểm định Pesaran cho thấy giá trị Pr = 0.0000 = 0.00% < 5%: Điều này cho thấy có sự phụ thuộc chéo giữa các biến Do đó, dữ liệu này cần áp dụng kiểm định tính dừng thế hệ thứ 2

Ho: Panels contain unit roots Number of panels = 114

Ha: Panels are stationary Number of periods = 13

AR parameter: Common Asymptotics: N/T -> 0

Panel means: Included

Time trend: Not included

ADF regressions: 1 lag

LR variance: Bartlett kernel, 7.00 lags average (chosen by LLC)

Ho: Panels contain unit roots Number of panels = 114

Ha: Panels are stationary Number of periods = 12

AR parameter: Common Asymptotics: N/T -> 0

Panel means: Included

Time trend: Not included

ADF regressions: 1 lag

LR variance: Bartlett kernel, 7.00 lags average (chosen by LLC)

Tương tự cho các biến khác

Ngoài kiểm định llc có thể thay thế bằng các kiểm định khác, tham khảo dưới đây:

xtunitroot ips varname [if] [in] [, IPS_options]

Fisher-type tests (combining p-values)

xtunitroot fisher varname [if] [in], {dfuller | pperron} lags(#)

[Fisher_options]

Hadri Lagrange multiplier stationarity test

xtunitroot hadri varname [if] [in] [, Hadri_options]

5.2 Kiểm định tính dừng thế hệ thứ 2 (Dữ liệu đang minh họa chọn loại này)

xtcips a,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for a

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

H0 (homogeneous non-stationary): bi = 0 for all i

Ta thấy giá trị trị tuyệt đối của CIPS là 0.901 nhỏ hơn giá trị trị tuyệt đối của Critical Value at 10%; 5%; 1% (đã lấy giá trị tuyệt đối tương ứng là 1.98; 2.04; 2.16) nên biến a không dừng Do đó, cần lấy sai phân bậc 1 & kiểm định tính dừng của biến sai phân bậc 1 của biến a

gen da=d.a

(114 missing values generated)

xtcips da,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for da

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

xtcips b,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for b

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

(114 missing values generated)

xtcips db,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for D.b

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

xtcips c,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for c

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

gen dc=d.c

(114 missing values generated)

xtcips dc,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for D.c

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

(228 missing values generated)

xtcips ddc,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for ddc

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

xtcips d,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for d

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

(114 missing values generated)

xtcips dd,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for dd

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

xtcips e,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for e

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

xtcips f,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for f

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

xtcips g,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for g

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

(114 missing values generated)

xtcips dg,bglags(1) maxlags(1)

Pesaran Panel Unit Root Test with cross-sectional and first difference mean included for dg

Deterministics chosen: constant

Dynamics: lags criterion decision General to Particular based on F joint test

Individual ti were truncated during the aggregation process

H0 (homogeneous non-stationary): bi = 0 for all i

Trên cơ sở đó, tổng hợp lại thành bảng như sau:

STT Biến Level (CIPS) First Different (CIPS) Two Different (CIPS) Kết luận

6 KIỂM ĐỊNH ĐỒNG LIÊN KẾT

6.1 Nếu dữ liệu có sự phụ thuộc chéo thì dùng kiểm định Westerlund (Dữ liệu đang minh họa chọn loại này)

xtcointtest westerlund a b c d e f g,allpanels

Westerlund test for cointegration

-

Ho: No cointegration Number of panels = 114

Ha: All panels are cointegrated Number of periods = 13

Cointegrating vector: Panel specific

Panel means: Included

Time trend: Not included

Ta thấy P-Value = 0.0657 = 6,57% > 5%: Nên không có đồng liên kết (Làm bước 7)

6.2 Nếu dữ liệu không có sự phụ thuộc chéo thì dùng kiểm định Pedroni hoặc Kao

xtcointtest pedroni a b c d e f g

Pedroni test for cointegration

-

Ho: No cointegration Number of panels = 114

Ha: All panels are cointegrated Number of periods = 12

Cointegrating vector: Panel specific

Panel means: Included Kernel: Bartlett

Time trend: Not included Lags: 2.00 (Newey-West)

AR parameter: Panel specific Augmented lags: 1

xtcointtest kao a b c d e f g

Kao test for cointegration

-

Ho: No cointegration Number of panels = 114

Ha: All panels are cointegrated Number of periods = 11

Cointegrating vector: Same

Panel means: Included Kernel: Bartlett

Time trend: Not included Lags: 1.44 (Newey-West)

AR parameter: Same Augmented lags: 1

White's test for Ho: homoskedasticity

against Ha: unrestricted heteroskedasticity

Fixed-effects (within) regression Number of obs = 1,254

Group variable: code Number of groups = 114

R-sq: Obs per group:

Random-effects GLS regression Number of obs = 1,254

Group variable: code Number of groups = 114

R-sq: Obs per group:

estimators for anything unexpected and possibly consider scaling your

variables so that the coefficients are on

b = consistent under Ho and Ha; obtained from xtreg

B = inconsistent under Ha, efficient under Ho; obtained from xtreg

Test: Ho: difference in coefficients not systematic

chi2(5) = (b-B)'[(V_b-V_B)^(-1)](b-B)

= 21.73

Prob>chi2 = 0.0006

xtreg da db ddc dd e f dg,fe

Fixed-effects (within) regression Number of obs = 1,254

Group variable: code Number of groups = 114

R-sq: Obs per group:

Modified Wald test for groupwise heteroskedasticity

in fixed effect regression model

H0: sigma(i)^2 = sigma^2 for all i

chi2 (114) = 1.4e+08

Prob>chi2 = 0.0000

xtreg da db ddc dd e f dg,fe

Fixed-effects (within) regression Number of obs = 1,254

Group variable: code Number of groups = 114

R-sq: Obs per group:

Cross-sectional time-series FGLS regression

Coefficients: generalized least squares

Panels: heteroskedastic

Correlation: no autocorrelation

Estimated covariances = 114 Number of obs = 1,254

Estimated autocorrelations = 0 Number of groups = 114

Estimated coefficients = 7 Time periods = 11

[-0.25] [1.40] [0.14] [-1.54]

dg 0.615 0.365 0.568 0.336* [1.16] [0.68] [1.08] [1.67]

_cons -0.132*** -0.349*** -0.140*** -0.0175** [-5.44] [-4.54] [-5.03] [-2.17]

-

N 1254 1254 1254 1254

R-sq 0.082 0.091

-

t statistics in brackets * p<0.1, ** p<0.05, *** p<0.01 Các mô hình tham khảo thêm: xtpcse da db ddc dd e f dg (Mô hình PCSE) Linear regression, correlated panels corrected standard errors (PCSEs) Group variable: code Number of obs = 1,254 Time variable: year Number of groups = 114

Panels: correlated (balanced) Obs per group: Autocorrelation: no autocorrelation min = 11

avg = 11

max = 11

Estimated covariances = 6555 R-squared = 0.0818 Estimated autocorrelations = 0 Wald chi2(6) = 26.22 Estimated coefficients = 7 Prob > chi2 = 0.0002 -

| Panel-corrected da | Coef Std Err z P>|z| [95% Conf Interval] -+ -

db | .0001477 .0000459 3.22 0.001 0000577 .0002377

ddc | -.1130193 .1970645 -0.57 0.566 -.4992586 .2732199

dd | .0094005 .0086826 1.08 0.279 -.007617 026418

e | .0006847 .0002296 2.98 0.003 0002347 .0011347

f | -.0001264 .0004246 -0.30 0.766 -.0009586 .0007059

dg | .6148719 .6815763 0.90 0.367 -.7209931 1.950737 _cons | -.1319026 .0267445 -4.93 0.000 -.1843207 -.0794844 -

ssc install xtabond2 (Cài đặt mô hình GMM)

checking xtabond2 consistency and verifying not already installed

all files already exist and are up to date

xtabond2 da db ddc dd e f dg,gmm(da, lag(1 1)) iv(db ddc dd e f dg) small two nocons (Mô hình GMM)

Favoring space over speed To switch, type or click on mata: mata set matafavor speed, perm

Warning: Two-step estimated covariance matrix of moments is singular

Using a generalized inverse to calculate optimal weighting matrix for two-step estimation

Difference-in-Sargan/Hansen statistics may be negative

Dynamic panel-data estimation, two-step system GMM

-

Group variable: code Number of obs = 1254

Time variable : year Number of groups = 114

Number of instruments = 27 Obs per group: min = 11

Warning: Uncorrected two-step standard errors are unreliable

Instruments for first differences equation

Arellano-Bond test for AR(1) in first differences: z = -2.78 Pr > z = 0.005

Arellano-Bond test for AR(2) in first differences: z = -1.15 Pr > z = 0.251

-

Sargan test of overid restrictions: chi2(21) = 732.87 Prob > chi2 = 0.000

(Not robust, but not weakened by many instruments.)